Introduction to Multivariate Data Analysis

Prof. Dr. Ursula Hoffmann-Lange

This course provides an introduction into two of the most basic and most widely used multivariate methods of data analysis, multiple regression and factor analysis. After a brief introduction into the importance of multivariate analysis for studying political phenomena both at the individual and the aggregate level, the course will cover the basic logic of these two methods, their underlying assumptions and how to deal with violations of these assumptions (e.g. multicollinearity). Students will be encouraged to develop their own research questions and to apply the methods in the analysis of general population surveys using SPSS.

This course provides an introduction in the basics of multivariate data analysis. Multiple ordinary least square (OLS) regression analysis and factor analysis are the two most widely used multivariate methods of data analysis in the social sciences. The course will provide a practical introduction into the underlying statistical models as well as their application in the analysis of comparative survey data, e.g. the European Social Survey (ESS), the International Social Survey Program (ISSP) or the World Values Survey (WVS). Students are expected to develop their own research questions and may also use their own data if they wish to do so. Basic knowledge of SPSS is helpful, but not essential.

The course will start out with a brief discussion of different types of variables (nominal, ordinal, intervall, ratio), types of statistical analysis (univariate, bivariate, multivariate) and the essentials of working with comparative data-sets (weights, recodes, index construction etc.). The introduction into multiple regression will focus on the basic logic and the central mathematical assumptions underlying the OLS regression model. Practical exercises will show how regression analysis can be used to test theoretical assumptions. The third part of the course will provide an introduction into exploratory factor analysis and its application for uncovering the structure and dimensionality of a set of items. While these two methods have been developed for the analysis of metric data, understanding their logic is a precondition for studying more complex methods for non-metric data.

Basic Literature:

  • G. David Garson, 2014: Multiple Regression. Asheboro (NC): G. David Garson and Statistical Associates Publishing.
  • G. David Garson, 2013: Factor Analysis. Asheboro (NC): G. David Garson and Statistical Associates Publishing.